sign in sign up
<<
Home , Long Valley Caldera

JOURNAL OF PETROLOGY VOLUME 38 NUMBER 12 PAGES 1661–1678 1997

Isotopic and Chemical Evolution of the Post- Rhyolitic System at Long Valley, California

ARND HEUMANN∗ AND GARETH R. DAVIES

FACULTEIT DER AARDWETENSCHAPPEN, VRIJE UNIVERSITEIT AMSTERDAM, DE BOELELAAN 1085, 1081 HV AMSTERDAM, THE NETHERLANDS

RECEIVED JANUARY 1997; ACCEPTED AUGUST 1997

Post-caldera of the Long Valley magmatic system are KEY WORDS: ; chamber; residence time; Sr–Nd–Pb chemically less evolved than pre-caldera rhyolites or the initial phases isotopes of the Bishop Tuff and record temporal variations in composition which imply open-system magma differentiation involving magma replenishment. All post-caldera rhyolites lie on a well-defined Pb–Sr isotope mixing line between the ~700 ka and the recent Inyo domes, precluding a simple cogenetic origin. Coherent INTRODUCTION temporal trends in Pb and Sr isotopes provide compelling evidence Variable volumes of highly differentiated rhyolitic for the near-continuous addition of magma into a silicic magma erupt periodically throughout the long lifetime chamber that was residual after Bishop Tuff eruption. Nd isotope of many large silicic magmatic systems. Volumetrically ratios do not record such consistent variations, arguing against their most important are the cataclysmic caldera-forming erup- use as a proxy for major additions of new magma and hence as a tions of ignimbrites derived from compositionally zoned monitor of potential volcanic hazard. The lowering of 87Sr/86Sr magma chambers. High-silica magmas, which evolve with time demonstrates that there was little crustal interaction and within the upper reaches of silicic magma chambers, are that the Nd isotope composition of the added magmas was variable. preferentially sampled before, or in the initial stages of, The rhyolites of Mammoth Knolls are the most differentiated and such cataclysmic events. Rhyolitic volcanism immediately define an Rb–Sr isochron of 277±124 ka. These data are after caldera collapse commonly marks a transient period consistent with stratification of the magma chamber and subsequent which leads to the onset of a new magmatic cycle. The isolation of the upper, more evolved, sections at ~275 ka. Post study of post-caldera volcanism may therefore provide 400 ka, rhyolites become chemically more varied, supporting evidence important information about the longevity and the rate of stratification. The high- and low-silica rhyolites record distinct of processes that regenerate high-silica derivatives in large temporal Sr isotope evolution, implying that the low-silica rhyolites silicic volcanic systems. sampled the bulk of the magma chamber whereas the high-silica Recent detailed isotopic studies of pre-caldera high- rhyolites represent the upper, isolated, parts of the magma chamber, silica rhyolites from the Long Valley magmatic system where they resided for >30 and <300 kyr. After the eruption of suggest that the conventional approach of monitoring the Bishop Tuff, the Long Valley magma chamber was well mixed the eruptive volumes of silicic magmatism through time for the first ~400 kyr of its evolution and produced rhyolitic magmas (e.g. Smith, 1979; Spera & Crisp, 1981) as a proxy for at an average of ~0·0001 km3/year. Upon stratification post 400 magma formation rates may be far too simplified and ka, tens of km3 of chemically more evolved magma were rapidly that rhyolitic magmas can form rapidly (Halliday et al., produced. Magma addition to the system was at a constant rate 1989; Halliday, 1990; Mahood, 1990; Davies et al., 1994; and more frequent than eruptions. Christensen & Halliday, 1996; Davies & Halliday, 1997).

∗Corresponding author. Telephone: +31-20-44-47403. Fax: +31-20- 646-2457. e-mail: [email protected]  Oxford University Press 1997 JOURNAL OF PETROLOGY VOLUME 38 NUMBER 12 DECEMBER 1997

Rhyolite production rates comparable with those of Tuff (Halliday et al., 1989; Metz & Mahood, 1991; basaltic systems (10–2 km3/year) and storage times of >100 Christensen & Halliday, 1996; Davies & Halliday, 1997). kyr have been inferred. Notwithstanding the thermal Subsequent to the eruption of the Bishop Tuff, the Long constraints imposed by the apparent longevity of such Valley magmatic system persisted to produce magmas of magma batches (e.g. Sparks et al., 1990), the almost predominantly rhyolitic composition. The post-caldera unique chemical signatures revealed in those studies are magmatic evolution of the Long Valley system has pre- difficult to reconcile with contrary petrogenetic models, viously been summarized by Bailey et al. (1976), Mankinen where rhyolites originate by rapid fusion of crustal et al. (1986), and Bailey (1989). After formation of the sources, followed by immediate ascent and eruption Long Valley caldera, early post-caldera rhyolites (>75 (Huppert & Sparks, 1988; Sparks et al., 1990). km3) were erupted between 750 and 650 ka forming a At Long Valley, rhyolitic volcanism resumed im- Resurgent Dome (Fig. 1). These aphyric to sparsely mediately after caldera formation at 760 ka (Pringle et porphyritic rhyolites contain only rare amounts of pla- al., 1992; van den Bogaard & Schirnick, 1995) and gioclase, orthopyroxene, biotite and Fe–Ti oxides, and has persisted episodically ever since. Several previous probably represent subaerial expressions of early grano- geochemical studies have been carried out on the post- phyric intrusions found within the intracaldera Bishop caldera magmatism within the caldera structure (e.g. Tuff (McConnell et al., 1995). Following formation of the Baranowski & Harmon, 1978; Bailey, 1984; Tanzer & Resurgent Dome and a quiescent period of ~100 kyr, Macdougall, 1984; Cousens, 1992), yet to date there are three groups of more crystal-rich moat rhyolites erupted no systematic combined chemical and isotopic data for at ~200 kyr intervals. Eruption started to the north of the post-caldera rhyolites. the resurgent dome and the locus of activity migrated In this study we present comprehensive major, trace clockwise around the Resurgent Dome (Fig. 1). Two element and isotope data for all post-caldera rhyolitic petrographically different types of moat rhyolites have units of the Long Valley system and evaluate the chemical been distinguished by Bailey (1989) mainly based on their evolution of the magma chamber with particular ref- phenocryst and silica contents: (1) low-silica rhyolites with erence to the timescale and rate of evolved magma a mineral assemblage of quartz, sanidine, plagioclase, generation. biotite and hornblende; (2) high-silica rhyolites with a similar mineral assemblage but lower mineral abund- ances. The youngest group of moat rhyolites (~100 ka) tem- GEOLOGICAL BACKGROUND porally and spatially overlaps basaltic and quartz latitic The Pliocene to Recent Long Valley magmatic system erupted between about 200 and 50 ka in the west evolved at the northern end of , California, moat of the caldera (Mankinen et al., 1986; Bailey, 1989; between the eastern front of the and the Cousens, 1992). These more mafic rocks follow a general western margin of the Basin and Range Province. An south–north trend of basaltic to intermediate lavas ex- outline of the regional geology and the stratigraphic tending from to , and relations among volcanic units, including references to are aligned with the Mono Craters and Inyo Craters the vast number of publications on the Long Valley area, (Fig. 1), the youngest expression of rhyolitic volcanism has been given by Bailey (1989). in the Long Valley region. The most prominent feature of the Long Valley mag- The Mono Craters chain consists of 30 overlapping matic system is the establishment of a large silicic magma rhyolitic domes located 10 km north of Long Valley chamber, which produced evolved rhyolites over the last caldera. The major volume of the volcanic chain erupted 2·1 Myr (Bailey et al., 1976). The magma chamber 1·2 and 0·6 kyr ago (Sieh & Bursik, 1986) from a relatively evolution culminated in the catastrophic eruption of the deep (10–20 km) magmatic source (Achauer et al., 1986; voluminous (>600 km3), chemically zoned Bishop Tuff Dawson et al., 1990), probably unrelated to the main (Hildreth, 1979) and consequent formation of the Long Long Valley magma chamber (Bailey et al., 1976; Bailey, Valley caldera at 760 ka (Pringle et al., 1992; van den 1984; Dawson et al., 1990). Bogaard & Schirnick, 1995). During the preceding 1·3 The most recent rhyolitic eruptions associated with the Myr period of pre-caldera activity, high-silica rhyolites caldera occurred coeval with the Mono Craters eruptions, (>50 km3) were episodically erupted at Glass Mountains, forming the Inyo Domes, a chain of mainly 0·6 kyr old north-east of the current caldera structure (Metz & Ma- domes and flows (Miller, 1985), which cut the north-west hood, 1985). These rhyolites compose a sequence of older margin of the caldera (Fig. 1). However, the origin of (2·1–1·2 Ma) and younger (1·2–0·79 Ma) lavas, which the Inyo magmas and their genetic relationship to the became compositionally more homogeneous and vo- Long Valley magmatic chamber remain controversial. luminous with time. The latest are in most respects Older Inyo Domes show chemical and petrographical chemically indistinguishable from early erupted Bishop similarities to the rhyolites of the Mono Craters (Bailey,

1662 HEUMANN AND DAVIES POST-CALDERA RHYOLITES, LONG VALLEY

Fig. 1. Geological map and sample locations of rhyolitic post-caldera units at Long Valley caldera [redrawn from Bailey (1989)].

1989), whereas the 0·6 ka Inyo flows comprise two SAMPLES AND ANALYTICAL complex commingled lava types: a coarsely porphyritic lava (CP), presumably coming from the Long Valley TECHNIQUES magma chamber, and a finely porphyritic lava (FP), The samples used in this study cover the entire suite of which itself was produced by post-crystallization mixing post-caldera rhyolites erupted within the margins of Long between rhyolitic and dacitic magmas apparently from Valley caldera (Fig. 1). We follow the terminology and the more northerly Mono Craters source (Bailey et al., geochronology of Bailey (1989) for the geological units. 1976; Sampson, 1987; Sampson & Cameron, 1987; Hig- We collected 14 and vitrophyres, rep- gins, 1988; Vogel et al., 1989; Gibson & Naney, 1992). resentative of flows that formed the Resurgent Dome Geophysical studies, designed to constrain the shape erupted between 750 and 650 ka (unit Qef ). The low- and size of the remnant silicic magma chamber at Long silica moat rhyolites (units Qmr1, -2, -3) comprise seven Valley caldera, have established that a relatively large felsophyre samples from the eruption interval 530–115 (150–600 km3) zone of molten or partially molten material ka. High-silica moat rhyolites include four samples of the still exists at 7–20 km depth beneath the Resurgent Dome rhyolites (unit Qmrh), erupted between 330 [see Dawson et al. (1990) and references therein]. These and 290 ka, and three samples of the Mammoth Knolls observations are in accordance with previously presented rhyolites (unit Qmrm), erupted between 110 and 97 ka. geological models (e.g. Bailey, 1984), which interpret the Sampling of the Inyo rhyolitic domes is difficult because post-caldera volcanism as being related to the cooling of the mingling of the compositionally different lavas. and successive solidification of a residual silicic magma However, we attempted to identify end-member com- chamber thermally sustained by addition of mafic magma positions at each locality and sampled the different facies at depth. (CP or FP lavas, respectively) from Deadman Creek

1663 JOURNAL OF PETROLOGY VOLUME 38 NUMBER 12 DECEMBER 1997

Dome (IC-46, -47), Glass Creek Dome (IC-49, -50, -51), RESULTS Dome (IC-52, -53), and the small dome north Major elements of Deadman Creek (IC-48). In this paper, our aim in sampling the Inyo Domes is simply to characterize their The post-caldera rhyolites define a large and continuous compositional range from low silica to high silica (72–77% isotopic signature and hence determine their relation- SiO ; Table 1, Fig. 2c). The Resurgent Dome forms the ship to the main magma volume at Long 2 major low-silica rhyolite (LSR) volume (~75 km3; 74–75% Valley. SiO2). The first moat rhyolites, which erupted to the NW After crushing of 6–9 kg of glass-rich samples, ~200 g of the Resurgent Dome, and the much younger Deer of washed rock chips were pulverized in agate mills. Mountain flow (IC-42) erupted in the south-west of the Whole-rock powders were analyzed on a PW 1404 X- caldera have the lowest SiO2 contents (<74%). The ray fluorescence (XRF) spectrometer using fused glass later moat rhyolites have SiO2 contents identical to beads for major elements and Ba concentrations, and the Resurgent Dome. High-silica rhyolites (HSR) are pressed powder pellets for trace elements (Sr, Rb, Y, Zr). confined to the south-east (Hot Creek rhyolites) and Additional trace elements (Nb, Cs, Hf, Ta, Th, U) and south-west (Mammoth Knolls rhyolites) of the caldera. rare earth element (REE) concentrations were determined All the rhyolites are metaluminous with an alumina by inductively coupled plasma-mass spectrometry (ICP- saturation index of ~1·0, and low TiO2 (<0·32%) and MS). International and internal standards were used P2O5 (<0·06%) contents (Table 1). Major element dia- as monitors of data quality. Reproducibilities of trace grams show linear relationships with decreasing Al2O3, elements by XRF and their agreement with ICP-MS CaO and TiO2 (Fig. 2a and b). None of the compositions results are better than 2% (Ba <10%). Reproducibilities are as extreme as the pre-caldera HSR of the Glass of ICP-MS results are better than 3% with the exception Mountain complex (Metz & Mahood, 1991) or early of Eu, Hf, Dy, U and Th (4–9%). For isotope analyses, phases of the Bishop Tuff (Hildreth, 1981). Rhyolites of 100–150 mg of rock powder were brought into solution. the Resurgent Dome have SiO2 contents comparable ff Pb, Sr, and Nd were sequentially separated by employing with those of later phases of the Bishop Tu (Hildreth, standard chromatographic ion-exchange techniques. 1981). Total blanks were equivalent to <0·01% and are con- sequently ignored. Isotopic compositions were measured on Finnigan Trace element abundances MAT 262 RPQ plus and MAT 261 multicollector ther- The post-caldera rhyolite suite has a larger range in some mal ionization mass spectrometers with eight movable trace elements than the pre-caldera Glass Mountain lavas and nine fixed collectors, respectively, operating in mul- and the Bishop Tuff. For example, Zr contents in the tidynamic (Sr, Nd) and static (Pb) modes. All errors post-caldera rhyolite (Fig. 2c–e) vary from 85 to 223 reported in the paper are 2 SE for within-run precisions ppm, in contrast to a range of 90–120 ppm for Glass and 2 SD for standards. During the course of analyses Mountain and 90–140 ppm for the Bishop Tuff (Hildreth, an internal Nd standard was monitored, which yielded 1979; Metz & Mahood, 1991). Although they have a an average 143Nd/144Nd of 0·511335±15 (n=25). The large range in some trace element contents (e.g. Ba, Sr), La Jolla standard gave 143Nd/144Nd=0·511859±9. NBS the post-caldera rhyolites are indicative of chemically less 987 yielded 87Sr/86Sr=0·710234±16 (n=36, MAT 262) evolved rocks relative to pre-caldera rhyolites. Sr contents and 0·710271±26 (n=76, MAT 261). All 87Sr/86Sr ratios are significantly higher than in pre-caldera lavas [39–237 of samples are normalized to 0·710250 for NBS 987. ppm compared with 0·1–2·5 ppm; see Davies et al. Fractionation was corrected to 146Nd/144Nd=0·7129 and (1994) for a summary]. Elements that behave the most 87Sr/86Sr=0·1194. Pb isotope ratios are normalized with incompatibly in this system (e.g. U, Th, Ta, Nb) have a mass discrimination factor of 0·117% per atomic mass large variations (e.g. Nb, 7–17 ppm) but these elements unit (a.m.u.) for NBS 981. do not approach the extreme values of the pre-caldera Because of the young age and the low Sm/Nd and lavas, where Nb, for example, reaches 45 ppm. U/Pb ratios, all initial Pb and Nd isotope ratios are within analytical precision of measured ratios. This is not true for the Sr isotope system, where age corrections Rare earth elements up to 0·00005 are required; however, errors in the initial The chondrite-normalized REE patterns of all the post- and measured isotope ratios are identical for all three caldera rhyolites have a similar shape with steep negative isotope systems. For age correction we used results from slopes for the light REE (LREE) and slightly positive previous K–Ar studies (Bailey et al., 1976; Mankinen et slopes for heavy REE (HREE) (Table 1, Fig. 3b). Eu al., 1986), except where noted in Table 2 (below). anomalies (Eu/Eu∗) vary from 0·6–0·7 in the Resurgent

1664 HEUMANN AND DAVIES POST-CALDERA RHYOLITES, LONG VALLEY Rhyolite of Mammoth 0·22 0·161·45 0·16 1·37 0·16 1·38 0·20 1·39 0·190·03 1·41 0·21 0·02 1·37 0·21 1·52 0·02 0·22 1·53 0·02 0·21 1·59 0·02 0·33 1·39 0·02 0·26 0·05 2·14 0·27 0·04 2·06 0·11 0·04 2·11 0·10 0·03 1·02 0·11 0·08 0·81 0·11 0·08 0·98 0·15 0·06 1·09 0·12 0·02 1·12 0·14 0·01 0·93 0·04 1·04 0·01 0·03 0·02 0·02 74·9 75·513·8 75·1 13·6 75·6 13·4 75·1 13·6 75·7 13·7 73·1 13·7 73·5 13·2 73·4 13·3 73·7 13·2 71·9 12·7 71·9 14·3 72·1 13·7 74·2 13·8 77·2 12·8 74·5 11·3 76·9 12·8 76·0 13·2 76·0 12·7 76·1 12·4 12·6 Resurgent Dome (Qef) Hornblende–biotite rhyolites (Qmr1, -2, -3) Rhyolite of Hot Creek (Qmrh) Knolls (Qmm) Table 1: Representative whole-rock analyses for post-caldera rhyolites of Long Valley T 3 3 2 5 O 3·68 3·91 3·78 3·90 3·75 3·81 3·58 3·76 3·77 3·24 4·23 3·74 3·97 3·51 2·97 3·49 4·07 3·77 3·79 3·84 2 O 2 O 2 2 O 5·22 5·20 5·14 5·20 5·19 5·25 4·66 4·66 4·63 4·83 4·43 4·71 4·59 5·20 4·62 5·11 4·99 4·52 4·60 4·64 O 2 2 P K Sample: IC-10v IC-09v IC-32v IC-21vSiO IC-36v IC-45v IC-31f IC-30f IC-29f IC-02f IC-42f IC-24f IC-38f IC-05f IC-03f IC-04f IC-01v IC-41f IC-40f IC-39f Al MnOMgOCaO 0·04Na 0·21 0·04 0·95 0·14 0·04 0·73LOI 0·14Total 0·04 0·70 0·15 0·03 100·97 0·33Ba 0·73 101·02 0·19Rb 0·25 0·03 100·95Sr 0·86 1246 0·16 0·04 101·24Y 0·98 151·2 0·79 0·30 995 101·07Zr 0·04 148·3 135·5 0·35 101·47 0·95Nb 0·31 1030 100·50 0·04 86·3Cs 135·2 16·4 0·48 100·77 1·00 203 0·34 1012 146·7 100·72 0·04 16·1 10·9 83·2 0·31 1·01 174 1131 100·64 0·32 138·5 0·05 10·8 2·82 5·7 101·08 14·7 85·5 0·91 1195 144·0 0·53 154 100·23 0·06 2·35 110·7 5·0 117·3 15·8 706 1·51 9·4 100·86 0·35 178 110·8 0·06 114·2 2·41 101·02 14·9 0·97 11·2 683 148·9 4·7 0·38 100·84 114·7 174 0·04 3·21 15·2 101·05 143·5 1·09 692 0·10 111·6 9·7 5·2 101·44 0·03 13·2 174 1·48 152·4 0·56 114·6 101·02 0·08 10·3 616 142·8 5·1 180 0·04 13·9 2·32 100·95 136·3 0·47 10·0 0·10 236·8 863 100·85 0·04 13·5 165 2·34 125·3 5·1 0·60 134·5 8·9 0·10 741 117·9 0·05 10·2 3·39 4·3 157 159·5 0·53 0·19 108·7 8·3 809 0·04 17·3 3·18 140 4·1 70·6 0·69 116·7 0·13 7·6 22·8 688 0·05 3·20 121·6 60·4 198 0·61 4·2 0·16 13·2 21·1 0·33 162·5 587 70·5 223 0·65 4·6 172·7 15·8 13·1 1·79 668 61·3 223 168·4 3·2 15·7 11·4 2·30 57·9 620 129 13·3 4·1 1·60 9·5 38·7 199 104 14·0 3·5 46·0 8·8 109 124 20·5 4·4 9·9 135 123 21·0 10·7 5·4 109 20·4 14·8 4·6 85 15·1 3·8 86 16·2 5·6 5·9 5·7 XRF TiO Fe

1665 JOURNAL OF PETROLOGY VOLUME 38 NUMBER 12 DECEMBER 1997 Rhyolite of Mammoth Resurgent Dome (Qef) Hornblende–biotite rhyolites (Qmr1, -2, -3) Rhyolite of Hot Creek (Qmrh) Knolls (Qmm) Table 1: continued

Sample: IC-10v IC-09v IC-32vICP-MS IC-21vLa IC-36vCe IC-45v IC-31fPr 42·5 IC-30f 71·3Nd 40·4 IC-29fSm 68·4 IC-02f 6·20 38·6 21·9Eu IC-42f 65·6 5·97 39·4 IC-24f 3·17Gd 20·8 IC-38f 67·1 5·80Tb 38·3 0·68 3·27 17·6 IC-05fDy 65·8 2·60 5·83 IC-03f 40·5 0·54 2·56 20·3Ho IC-04f 0·39 68·7 2·61 5·79 40·2 0·48 2·93 17·5Er IC-01v 2·25 0·39 67·4 2·56 IC-41f 6·07 37·0Tm 0·53 2·69 0·52 18·6 2·17 IC-40f 0·40 61·9 2·44Yb 5·84 36·3 0·56 1·64 IC-39f 2·76 0·51 17·1 1·97Lu 0·30 0·35 61·6 2·40 5·65 32·5 0·58 1·58 2·55 0·45 16·9Hf 2·29 1·87 0·29 0·42 52·8 2·59 5·59 47·6 0·44 1·47 2·64Ta 0·48 16·8 0·29 2·00 1·82 0·25 0·44 84·9 2·25 4·44Th 37·3 0·43 1·57 2·60 0·45 4·85 12·6 0·29 2·06 1·55 0·28 0·39U 69·6 2·23 7·60 1·48 40·8 0·44 1·53 1·81 0·48 4·47 24·1 0·25 1·77 16·4 1·73 0·25 0·40 78·3 2·24 7·17 1·52 38·4 0·38 1·60 3·52 0·43 4·14 23·2Sample 0·27 1·90 codes: 16·3 1·66 0·26 v, 0·39 5·85 64·5 vitrophyre; 1·80 7·65 1·46 o, 34·1 0·73 1·29 obsidian; 4·12 0·44 4·47 24·5 f, 0·25 1·86 felsophyre. 15·2 1·59 0·22 0·27 6·00 57·7 3·07 5·65 1·49 39·3 0·64 1·47 3·98 0·43 4·53 16·4 0·28 1·27 16·4 1·47 0·23 0·53 5·61 67·0 3·77 5·05 1·47 40·4 0·73 1·40 2·25 0·30 4·68 14·3 0·24 2·39 16·1 1·44 0·24 0·63 6·03 68·5 3·72 5·79 1·53 22·6 0·32 1·03 2·10 0·56 4·81 16·6 0·24 3·08 16·3 1·42 0·18 0·66 6·03 45·8 2·31 5·90 1·39 21·7 0·28 1·73 2·43 0·71 4·47 17·2 0·24 2·91 14·4 1·15 0·30 0·38 6·09 43·7 2·06 4·57 1·38 22·9 0·33 2·46 2·57 0·66 4·07 14·9 0·19 1·71 14·0 1·78 0·38 0·32 5·27 46·8 2·15 4·41 1·39 0·30 2·22 2·98 0·39 3·70 14·3 0·29 1·50 13·7 2·32 0·37 0·38 5·28 2·33 4·62 1·46 0·28 1·35 2·91 0·36 5·02 15·0 0·37 1·77 13·8 2·25 0·24 0·41 5·11 2·85 1·71 0·21 1·27 2·84 0·42 5·77 0·34 1·78 12·7 1·35 0·19 0·54 4·70 2·82 2·28 0·25 1·33 0·42 5·81 0·25 2·57 15·3 1·23 0·22 0·54 4·12 2·68 2·00 1·40 0·62 3·62 0·20 2·56 14·5 1·44 0·26 0·55 5·05 1·62 2·10 0·61 3·16 0·22 2·70 14·0 1·53 0·35 4·84 1·46 2·14 0·63 3·73 0·25 12·9 2·27 0·38 4·85 1·62 2·09 3·51 0·36 14·3 2·31 0·38 4·49 1·75 3·65 0·36 15·0 2·20 4·72 2·64 3·18 0·38 18·6 4·88 2·71 3·36 18·9 6·88 2·79 19·2 7·15 7·37

1666 HEUMANN AND DAVIES POST-CALDERA RHYOLITES, LONG VALLEY

Fig. 2. Variations of selected major and trace element concentrations and ratios for post-caldera rhyolites from Long Valley caldera. It should be noted that the major elements define coherent differentiation trends for the entire suite. In contrast, REE and high field strength elements (HFSE) record two fractionation paths. Vectors show the nature and magnitude to which different minerals will fractionate REE and HFSE. Partition coefficients from Arth (1976), Mahood & Hildreth (1983) and Sisson (1994).

1667 JOURNAL OF PETROLOGY VOLUME 38 NUMBER 12 DECEMBER 1997

Dome and LSR to 0·2 in the most evolved HSR (Fig. DISCUSSION 2g). There are significant variations within the REE Chemical evolution content and REE patterns within single stratigraphic units (Table 1). In general, the oldest post-caldera rhyo- Compositional trends lites from the Resurgent Dome are less variable than the The major and trace element data presented in the younger moat rhyolites, with the LSR being characterized previous section demonstrate that the post-caldera rhy- by a wide range in HREE and having comparatively olites at Long Valley have a large compositional range. elevated LREE. The high-silica rhyolites have a marked The aim of the following discussion is to establish to negative Eu anomaly and slightly lower LREE contents. what extent this suite represents cogenetic magmas and Compared with the highly evolved high-silica rhyolites to place constraints on the temporal evolution of the of Glass Mountain and the early Bishop Tuff, which Long Valley magma chamber following eruption of the ff have relatively flat REE patterns (Fig. 3a; Hildreth, 1981; Bishop Tu . Metz & Mahood, 1991), post-caldera rhyolites exhibit Estimating the extent of fractional crystallization in less evolved REE characteristics, greater LREE abund- high-silica rhyolites using trace elements has been shown to be difficult because of the extreme compositional ances and enrichment, and lower HREE contents (Fig. dependence of partition coefficients and the difficulty in 3b). Resurgent Dome rhyolites have REE contents gen- determining starting compositions (e.g. Hildreth, 1979, erally similar to those of the late phases of the Bishop 1983; Mahood & Hildreth, 1983; Michael, 1983). Fur- Tuff. thermore, compositional zonations in silicic magma chambers can be misleading so that models will be influenced by which parts of a magma chamber are erupted. The post-caldera rhyolites of this study reveal a con- Sr–Pb–Nd isotope data tinuous compositional range in their major elements, The Rb and Sr concentrations of the post-caldera rhy- which could indicate that fractional crystallization is the olites range from 109 to 173 ppm and from 39 to 237 dominant process governing the evolution from low- to ppm, respectively (Table 2). 87Rb/86Sr ratios are therefore high-silica rhyolite. The HSR are characterized by large ∗ much less extreme (1·34–13·5) than in the highly evolved negative Eu anomalies (Eu/Eu >0·2; Figs 2b and 3). pre-caldera rhyolites of Glass Mountains or the early These large Eu anomalies are associated with low Sr Bishop Tuff, which reach 11 000 (e.g. Halliday et al., contents (>30 ppm) and are too extreme to be explained 1989; Davies et al., 1994). Present-day 87Sr/86Sr and by any crustal melting model (e.g. Halliday et al., 1991). 143Nd/144Nd (Table 2) are distinct in the different rhyolite In addition, the Nd isotope ratios of the rhyolites are too units. The Resurgent Dome records the highest 87Sr/86Sr radiogenic to be derived by melting local crustal material (e.g. Halliday et al., 1989; Davies et al., 1994). Con- (0·706592–0·706767) and lowest e (–2·38 to –2·16); Nd sequently, the Eu anomalies are evidence of extreme low-silica moat rhyolites have lower 87Sr/86Sr (0·706124– feldspar fractionation. There is a good linear correlation 0·706430) and more radiogenic e (–1·44 to –1·31). Nd between Zr and Eu/Eu∗ (Fig. 2g), which implies that High-silica rhyolites have relatively high 87Sr/86Sr feldspar and zircon crystallized throughout the evolution (0·706507–0·706351) and e (–1·87 to –0·74). The iso- Nd of the system. The decrease in Zr from 225 ppm in the tope data of the dominantly low-silica Inyo Domes con- LSR to ~100 ppm in the HSR and 80 ppm in the tinue this compositional dependence of the isotope Mammoth Knolls samples (Fig. 2c) requires <1% zircon compositions with less radiogenic signatures for Sr and fractionation. 87 86 = more radiogenic for Nd ( Sr/ Sr 0·706034–0·706195; The mineral vectors in Fig. 2 demonstrate that small e = Nd –1·70 to 0·51). The Nd isotope values lie in the amounts of accessory minerals, such as zircon, allanite ff range recorded by the Bishop Tu and pre-caldera Glass or apatite, can have enormous leverage on the shape of Mountain lavas (Halliday et al., 1984, 1989; Davies et al., REE patterns in rhyolitic magmas. Only few mineral 1994; Christensen & Halliday, 1996; Davies & Halliday, phases potentially fractionate the La/Sm ratio of magmas 1997). (e.g. allanite, pyroxenes, amphiboles). The bulk of the 208Pb/204Pb and 207Pb/204Pb ratios in the post-caldera REE and trace element variations in Fig. 2d, f and h could rhyolites are indistinguishable within the accepted ana- be modelled by combined zircon–apatite fractionation. lytical precision. 206Pb/204Pb ratios, however, record a However, the inferred 1% of apatite required to explain significant decrease with time from 19·30 to 19·11. Over- the La/Sm variation is not realistic and requires other all the Pb isotope compositions are similar to values minerals, such as hornblende, which fractionates REE previously reported from the pre-caldera lavas (Halliday in a similar manner to apatite. Previous studies have et al., 1989; Davies et al., 1994). shown that mineral partition coefficients of major mafic

1668 HEUMANN AND DAVIES POST-CALDERA RHYOLITES, LONG VALLEY

Fig. 3. Chondrite-normalized REE abundances of (a) post-caldera rhyolites and the pre-caldera Glass Mountain and Bishop Tuff systems (Metz & Mahood, 1991), and (b) post-caldera rhyolites from Long Valley caldera.

phases (pyroxenes and amphiboles) can be strongly in- With the exception of its influence on the Eu anomaly, fluenced by accessory mineral inclusions enriched in major feldspar fractionation would result in only minor LREE (Cameron & Cameron, 1986; Michael, 1988). LREE fractionation (Nash & Crecraft, 1985). Con- Recent ion microprobe studies on hornblende, including sequently, the majority of REE fractionation of the post- rhyolite samples from the Long Valley system, have, caldera rhyolites can be explained by hornblende, zircon however, confirmed high mineral–melt partitioning for and ±apatite, minerals that are among the observed the REE in amphiboles (Sisson, 1994). The increasing phenocryst assemblage. Absolute REE abundances La/Sm ratio in lavas erupted between 700 and 300 ka (Table 1) and inter-REE diagrams (La/Sm vs La/Yb, (Fig. 5b, below), therefore can also be explained by Sm, Yb, La) yield coherent trends which can be modelled ~5% fractionation of hornblende (La/Sm ~0·1). This by 5% hornblende and <0·1% accessory zircon frac- percentage exceeds the amount of mafic minerals present tionation (e.g. Fig. 2f and h). Before 300 ka, LREE- in the rocks, but does not exclude a control of La/Sm enriched minor phases cannot be involved in the frac- ratios by hornblende and apatite in the genetic evolution tionation assemblage, because of the consistent increase of the rhyolites. in La/Sm and La/Yb ratio with fractionation (SiO2) Other major phenocryst phases (quartz and feldspars) associated with little change in total LREE should have little effect on the slope of the REE pattern. abundances.

1669 JOURNAL OF PETROLOGY VOLUME 38 NUMBER 12 DECEMBER 1997

Table 2: Sr, Nd and Pb isotope data for post-caldera rhyolites of Long Valley

87 86 87 86 206 204 207 204 208 204 143 144 Unit Sample∗ Age (ka)† Rb/Sr Sr/ Sr Sr/ Sri Pb/ Pb Pb/ Pb Pb/ Pb Nd/ Nd eNd

Resurgent Dome Qef IC-10v 751±16 1·08 0·706767±10 0·706733 19·250 15·663 38·908 0·512516±7 −2·38 Qef IC-37o 694±17 1·26 0·706683±9 0·706647 19·271 15·678 38·964 Qef IC-22o 692±14 1·71 0·706612±9 0·706563 19·236 15·666 38·877 0·512541±4 −1·89 Qef IC-06v 688±14 1·32 0·706723±9 0·706686 19·298 15·699 39·028 0·512523±5 −2·24 Qef IC-08v 680±40 1·69 0·706636±9 0·706589 19·264 15·673 38·927 Qef IC-09v 680±29 1·70 0·706592±9 0·706545 19·254 15·672 38·928 0·512529±15 −2·16 Qef IC-12o 680±40 1·28 0·706657±9 0·706621 19·237 15·660 38·872 Qef IC-15v 680±40 1·36 0·706669±8 0·706631 19·260 15·672 38·905 Qef IC-16v 680±40 1·74 0·706603±8 0·706554 19·242 15·661 38·881 Qef IC-25o 680±40 1·63 0·706593±8 0·706547 19·228 15·664 38·879 Qef IC-32v 680±40 1·62 0·706681±8 0·706636 19·213 15·670 38·873 0·512546±10 −1·79 Qef IC-35v 680±40 1·29 0·706662±9 0·706626 19·279 15·695 39·016 0·512520±5 −2·30 Qef IC-18v 678±28 1·36 0·706678±9 0·706640 19·241 15·678 38·930 Qef IC-20v 652±14 1·74 0·706615±8 0·706568 19·265 15·684 38·970 0·512542±8 −1·87 Hornblende–biotite rhyolites Qmr1 IC-31f 527±12 0·81 0·706419±9 0·706401 19·232 15·691 38·954 0·512561±10 −1·50 Qmr1 IC-30f 523±11 0·81 0·706407±8 0·706389 19·252 15·703 39·002 0·512580±11 −1·13 Qmr1 IC-29f 481±10 0·77 0·706430±8 0·706415 19·227 15·682 38·927 0·512539±11 −1·93 Qmr2 IC-02f 362±8 0·78 0·706304±9 0·706292 19·206 15·671 38·875 0·512571±7 −1·31 Qmr3 IC-42f 115±3 0·49 0·706301±18 0·706299 19·191 15·696 38·959 0·512548±9 −1·76 Qmr3 IC-24f 152±2‡ 1·04 0·706140±17 0·706135 19·175 15·682 38·921 0·512572±8 −1·29 Qmr3 IC-38f 148±2‡ 0·80 0·706124±14 0·706120 19·159 15·678 38·896 0·512564±5 −1·44 Rhyolite of Hot Creek Qmrh IC-05f 329±23 1·69 0·706445±14 0·706422 19·216 15·686 38·922 0·512578±10 −1·17 Qmrh IC-03f 300±30 1·82 0·706476±13 0·706454 19·245 15·728 39·048 0·512600±7 −0·74 Qmrh IC-04f 300±30 1·71 0·706507±9 0·706486 19·245 15·717 39·027 0·512589±11 −0·96 Qmrh IC-01v 288±31 2·01 0·706414±9 0·706390 19·212 15·687 38·930 0·512582±7 −1·09 Rhyolite of Mammoth Knolls Qmrm IC-41f 110±11 2·87 0·706330±13 0·706317 19·182 15·687 38·930 0·512565±10 −1·42 Qmrm IC-40f 106±2 4·67 0·706351±13 0·706331 19·178 15·689 38·928 0·512563±11 −1·46 Qmrm IC-39f 97±6 3·70 0·706330±15 0·706315 19·178 15·690 38·932 0·512542±15 −1·87 Inyo Domes Qri IC-46f ~0·6 0·89 0·706063±10 0·706063 19·136 15·674 38·867 Qri IC-47v ~0·6 0·98 0·706050±9 0·706050 19·142 15·683 38·902 0·512551±7 −1·70 Qri IC-49f ~0·6 0·50 0·706195±9 0·706195 19·154 15·667 38·846 Qri IC-50f ~0·6 0·46 0·706192±9 0·706192 19·162 15·679 38·887 0·512564±8 −1·44 Qri IC-51v ~0·6 1·22 0·706034±9 0·706034 19·123 15·678 38·882 Qri IC-52f ~0·6 3·51 0·706077±10 0·706077 19·120 15·677 38·892 0·512612±7 −0·51 Qri IC-53v ~0·6 3·14 0·706090±9 0·706090 19·125 15·673 38·895 Qro IC-48f ~0·6 6·39 0·706073±10 0·706073 19·115 15·685 38·911 0·512592±12 −0·90

∗Sample codes: v, vitrophyre; o, obsidian; f, felsophyre. †Ages from Bailey et al. (1976) and Mankinen et al. (1986); 680±40 is the estimated mean age for Resurgent Dome. ‡40Ar–39Ar ages of sanidines (Heumann, unpublished data).

1670 HEUMANN AND DAVIES POST-CALDERA RHYOLITES, LONG VALLEY

The rhyolites of Mammoth Knolls (~100 ka) have different trace element and REE characteristics compared with other post-caldera rhyolites (e.g. LaN ~70 compared with >100). Interestingly, however, they lie at the end of the Zr vs Eu/Eu∗ or Rb/Sr fractionation trends (Fig. 2e and g) implying that they have undergone the most extensive fractionation. This conclusion is consistent with their more evolved major element composition, but they are also characterized by higher Th, Ta, Nb and HREE contents than the other post-caldera rhyolites. Because of the extreme LREE partition coefficients of minor phases such as allanite and monazite in highly silicic systems (Evans & Hanson, 1993), as little as 0·03% fractional crystallization of these phases (Fig. 5d, f and h; below) would explain the observed relative LREE depletion (assuming partition coefficients of allanites from the Bishop Tuff; Mahood & Hildreth, 1983). This small degree of crystallization is insufficient to produce sig- nificant depletion in highly incompatible elements such as Nb and Ta. Furthermore, allanite is, next to possibly biotite, the only cogenetic mineral that can lower the La/Sm ratio. Zircon saturation temperatures for the Mammoth Knolls are the lowest of all post-caldera rhy- olites, at the end of a decreasing temperature trend from 780°Cto720°C over a period of 500 kyr (Fig. 5a, below). This observation supports control of the LREE by allanite, similar to the strong temperature dependence ff on the occurrence of allanite in the Bishop Tu (Mahood Fig. 4. Initial 87Sr/86Sr vs (a) 143Nd/144Nd and (b) 206Pb/204Pb for post- & Hildreth, 1983). Therefore the involvement of ac- caldera rhyolites from Long Valley caldera. cessory phases in the petrogenesis of these rocks is likely. The Mammoth Knolls lavas have many similar chemical characteristics to the entire post-caldera rhyolitic suite, Isotopic variations but REE ratios suggest a different fractional crystallization The Long Valley post-caldera rhyolites have a relatively large range in Sr, Nd and Pb isotope ratios. For example, history and isolation from the bulk of the magma system. 87 86 In terms of trace element contents, there is a coherent Sr/ Sr ranges from 0·70604 in the Inyo Domes to temporal evolution of the rhyolite system. With the 0·70677 in the Resurgent Dome. Each geographic group has a spread in values significantly outside analytical exception of the most recent LSR, there is a gradual error (Fig. 4). Despite this isotopic variability, the entire depletion in Zr from 730 ka (the Resurgent Dome) to rhyolite suite defines a coherent linear array on a Pb–Sr 97 ka (the Mammoth Knolls). Over the same timescale, isotope diagram (Fig. 4b) between the radiogenic values Sr contents in the HSR become lower, which is in of the Resurgent Dome and the relatively unradiogenic contrast to the LSR and their increased Sr contents values of the Inyo lavas. (Table 1). Most trace elements behave compatibly over Several possible processes could have produced this the first 400 kyr (Fig. 5b, d, f and h; below) before significant Pb–Sr isotope variation. First, magmas par- increasing to the values of the Mammoth Knolls, which ental to the post-caldera rhyolites could have progressively are, however, not as enriched as the pre-caldera Glass undergone less crustal interaction with time before en- ff Mountain lavas or early phases of the Bishop Tu (Fig. tering the magma chamber. Second, melting of the lower 5d and e, below). This implies that either two distinct crust because of basaltic underplating (Hildreth, 1981; magma systems (HSR and LSR) exist or that at ~400 ka Huppert & Sparks, 1988) could be a disequilibrium batches of high-silica rhyolites were effectively separated process, with initial melts being more radiogenic as a from a low-silica volume, because of, for example, the result of preferential biotite breakdown and subsequent stratification of the chamber. The coherence of the major melts being more controlled by feldspars (Tommasini and trace element variations would appear to favour the & Davies, 1997). These models, however, appear very second alternative. arbitrary, particularly if the long-lived nature of the silicic

1671 JOURNAL OF PETROLOGY VOLUME 38 NUMBER 12 DECEMBER 1997

Fig. 5. Temporal variation of Sr–Nd–Pb isotopes and trace elements (Nb, Yb) in post-caldera rhyolites. Continuous lines in (b), (d), (f ) and (h) highlight the temporal evolution in trace element contents and ratios with a marked change in slope at ~400 ka. In (g), the dashed line indicates how the Resurgent Dome and low-silica rhyolites define a consistent reduction in initial 87Sr/86Sr of ~9×10–6 per 10 kyr. Arrows indicate how high-silica rhyolites can be compared with their parental source before extended isolation at the top of the magma chamber (see text for detailed discussion).

1672 HEUMANN AND DAVIES POST-CALDERA RHYOLITES, LONG VALLEY

system is taken into account. In addition, >600 km3 of Mixing silicic magma had already been produced in the area, The linear correlation defined by the Pb–Sr isotope much of which has lower initial Sr isotope ratios than ratios of the post-caldera lavas, between Resurgent Dome the post-caldera magmas, so that it is difficult to argue magma and later erupted lavas, places stringent con- that the relatively small volume of the more recent straints on the compositional nature of the magma added magmatism caused significant changes in the physico- to the system. The shape of mixing curves on a Pb–Sr chemical structure of the crust. The favoured in- isotope diagram is a function of the concentrations and terpretation of the isotope data is that the rhyolite suite isotopic compositions of the respective end-members. represents the product of a single system formed by Binary mixing calculations (Fig. 6) indicate that the new the mixture of magmas from two distinct sources, both magma must have Pb/Sr ratios very similar to the isotopically homogeneous in terms of Pb and Sr; namely, residual magma which formed the Resurgent Dome, but magma residual after the Bishop Tuff eruption and a significantly lower 206Pb/204Pb and 87Sr/86Sr isotope ratios. new magma input. The Pb isotope ratios define a strong The Pb isotope ratios of pre-caldera mafic and inter- temporal relationship, with the 206Pb/204Pb ratios de- mediate lavas (Nielsen et al., 1991) are the lowest values creasing from 19·30 in the Resurgent Dome to 19·11 in for magmatic rocks in the area and were chosen as the Inyo lavas (Fig. 5c). This observation rules out the isotopic composition of the second magma. Hypothetical entire rhyolite suite being cogenetic and produced by mixing curves for this scenario were calculated assuming fractional crystallization. The isotopic data imply an open anti-correlated Sr and Pb concentrations. A good fit can system with semi-continuous addition of material to the only be achieved with Sr/Pb ratios of 10–20. High Pb magma chamber with less radiogenic Pb and Sr isotope concentrations (15–20 ppm) and Sr between 200 and ratios (Fig. 4b). Because of the radiogenic Sr and Pb 300 ppm are required for realistic amounts of magma isotope composition of the crust in the Long Valley addition. Higher Sr concentrations are permissible (up region [see Davies et al. (1994) for a review] the coupled to 500 ppm), but then at least 30 ppm of Pb is required decrease in Pb and Sr isotope ratios with time cannot to attain a low hyperbolic curvature. In the preferred be ascribed to upper-crustal interaction of the magma scenario (curve D or E, Fig. 6), this would require a while residing in a magma chamber. In addition, the magma recharge of ~40% to the system. Although the vast majority of post-caldera lavas have low Rb/Sr ratios inferred Sr concentrations can be found in rhyolites and so that the production of radiogenic Sr from in situ less evolved or basalts, high Pb concentrations can only be expected in already highly fractionated radioactive decay in the magma chamber will be small, magmas, such as rhyolites. With the Resurgent Dome in marked contrast to the pre-caldera lavas, which have magma as the probable end-member, none of the later Rb/Sr ratios up to 3600 (Davies et al., 1994). We therefore erupted rhyolites at Long Valley have the major or trace conclude that the Pb–Sr isotope relationship is recording element composition required to fit observed isotope– the progressive introduction of more juvenile magmas trace element mixing curves. In addition, post-caldera from depth to a magma body that survived eruption of ff basaltic lavas (Cousens, 1992), erupted from ~260 ka the Bishop Tu . onwards, have 87Sr/86Sr (0·70536–0·70634) and 206Pb/ The coherent coupling of isotope ratios is much less 204Pb (19·18–19·27) that are too high to serve as mixing pronounced for Nd isotope data from the post-caldera end-members. Moreover, their different trace element rhyolites. However, there is a general correlation, in- contents and Pb/Sr ratios produce mixing hyperbolas e dicating higher Nd with time (Fig. 5a) and decreasing with large curvatures. 87 86 initial Sr/ Sr (Fig. 4a). This temporal evolution in 87 86 In Fig. 7a, the relationship between Sr/ Sri and 1/Sr Nd–Sr–Pb isotopes implies progressive incorporation of demonstrates that for the Resurgent Dome the decrease of a magmatic component with more mantle-like Nd isotope 87 86 Sr/ Sri with time is correlated with Sr concentration. signatures. Similar conclusions have been drawn from However, the very similar Pb/Sr ratios required for the previous isotope studies on the pre-caldera rhyolites and near-linear Sr–Pb correlation, coupled with the temporal the Bishop Tuff (Halliday et al., 1989; Davies et al., 1994), isotopic variations and the large differences in major which established a change in Nd isotope compositions element composition of contemporaneous magmas (e.g. e at ~1·2 Ma with Nd from –3·9 to –2·6 before 1·2 Ma HSR and LSR), argue against a mixing process being and from –1·2 to –0·5 thereafter. The contrast between solely responsible for the compositional variations. The the coherent nature of the temporal variations in Sr–Pb major and trace element trends outlined in the previous isotope ratios of post-caldera rhyolites and the greater sections lend support for a fractionation process to explain variation of the Nd isotope data implies that the magmas the variations in Sr concentrations in combination with added to the system did not have a constant Nd isotope a temporal decrease of Sr isotope compositions. Con- ratio. sequently, to satisfy the demands imposed by the Pb–Sr

1673 JOURNAL OF PETROLOGY VOLUME 38 NUMBER 12 DECEMBER 1997

Fig. 6. Initial 87Sr/86Sr vs 206Pb/204Pb diagram with hypothetical mixing curves between a Resurgent Dome end-member A and end-member B which has an isotopic composition of the pre-caldera mafic and intermediate lavas (Nielsen et al., 1991). Curves a–g represent mixing hyperbolas with Sr/Pb ratios varied between 600 and 1·67 (see text for detailed discussion). isotope relationships and the apparent inadequacies of a system is provided by the Rb–Sr isotope systematics of simple binary mixing, a compositionally similar magma lavas from the Resurgent Dome. They record a negative with different isotopic ratios must have been added to slope on an Rb–Sr isochron diagram (Table 2), implying the system more or less independently from the processes that less radiogenic magma had been introduced to that caused the compositional variations in major and the magma chamber to mix with material remaining trace elements amongst post-caldera rhyolites. Hence a immediately after the caldera-forming eruption of the model more complex than two-component mixing is Bishop Tuff. The implied mixing relationship requires required to explain the post-caldera evolution of the addition of a magma with a relatively high Rb/Sr ratio system. (>2) and an initial Sr isotope ratio <0·7065. It appears highly unlikely that a partial melt from the crust would have these characteristics. Therefore it is probable that this magma had undergone significant fractional crys- Origin of high-silica rhyolites tallization to produce a moderate Rb/Sr ratio before The temporal evolution of the Sr isotope ratios records mixing with magmas that remained in the chamber after a distinction between the HSR and LSR. The LSR eruption of the Bishop Tuff. have initial 87Sr/86Sr ratios that are ~0·0002 lower than Previous isotope studies have suggested extended pre- approximately contemporaneous HSR (Fig. 5b). Despite eruptive storage of magma for the youngest post-caldera the coherent major and trace element variations (Figs 2 rhyolites. For the FP series of the Inyo Domes, Sampson and 4a), this Sr isotope difference could be used to argue et al. (1984) estimated, on the basis of 238U–230Th dis- for derivation of the HSR and LSR from different sources. equilibrium, a fractionation event between 6 and 100 ka. The coupled temporal variations in Pb–Sr isotope ratios, Hornblende and allanite in CP lavas appear to have however, argue against different source regions for LSR formed 20 kyr before eruption (Sampson & Cameron, and HSR, and imply that the two suites are derived from 1987). Zircon crystallization ages (~200 ka) from LSR different parts of a single evolving magma system. and HSR from Mammoth Knolls and Deer Mountain Further evidence for progressive addition of magma appear to precede eruption ages by ~100 kyr (Reid et with relatively unradiogenic Sr to the Long Valley magma al., 1997). Unfortunately, few of the post-caldera rhyolites

1674 HEUMANN AND DAVIES POST-CALDERA RHYOLITES, LONG VALLEY

from the initial ratios defined by the regional isochrons of the pre-caldera rhyolites (Halliday et al., 1989; Davies et al., 1994; Davies & Halliday, 1997). The lavas which define the isochron were erupted over an ~18 kyr period (115±3kato97±6 ka; Bailey et al., 1976; Mankinen et al., 1986). The age defined by the Rb–Sr isochron is outside the uncertainty of the eruption age of the lavas. The difference between the eruption ages and the Rb–Sr isochron implies a minimum magma residence time of 35 kyr and a maximum close to 200 kyr. The latter time approaches the values obtained from the pre-caldera Glass Mountain rhyolites (Davies et al., 1994; Davies & Halliday, 1997). Although Rb–Sr whole-rock isochrons have potential problems arising from magma mixing, it has to be noted that the HSR are homogeneous glass (>80%) and that plots of initial 87Sr/86Sr vs 1/Sr do not show good linear correlation (r 2=0·62). Despite the general mixing trend documented in the Pb–Sr isotope relationship of all post- caldera rhyolites over the past 600 kyr, the Mammoth Knolls rhyolites record no petrologic, isotopic or chemical mixing, in contrast to the mingling of the Inyo lavas (e.g. Sampson & Cameron, 1987; Higgins, 1988; Vogel et al., 1989; A. Heumann, unpublished data, 1997). Moreover, an internal Rb–Sr isochron on minerals and glass of the dome with the highest Rb/Sr ratio yields an identical age, and initial results from a 230Th–238U disequilibrium investigation lend support of extended magma storage prior to eruption (Heumann et al., in preparation). These data therefore appear indicative of significant Fig. 7. (a) Initial 87Sr/86Sr vs 1/Sr for post-caldera rhyolites. Lavas pre-eruption formation of magmas comparable with re- from the Resurgent Dome define a linear relationship. (b) Rb–Sr cent results from other silicic systems throughout the isochron diagram for rhyolites of Mammoth Knolls and Deer Mountain. world (Sampson et al., 1984; Halliday et al., 1989; Chris- tensen & DePaolo, 1993; Davies et al., 1994; Heumann et al., 1995; Dunlap & Gill, 1997; Reid et al., 1997). have Rb/Sr ratios high enough to allow an assessment These data also provide important constraints upon the of their magma residence times. The exceptions are the physical nature of the magma system. To preserve a pre- youngest HSR (~100 ka) in the west moat of the caldera eruptive isochron, the different magma batches must at Mammoth Knolls, which have Rb/Sr ratios up to 4·7. have been formed from magma(s) having identical initial Three domes (~0·1 km3) were erupted in this region Sr isotope ratios and subsequently have been isolated. along with a large LSR lava flow (<0·6 km3) and the Mixing within the chamber and significant fractional small LSR of Deer Mountain (IC-42). The crystallization would disrupt any isochronous re- Deer Mountain dome has Pb and Nd isotope ratios that lationships [see Halliday et al. (1989); Halliday (1990) and are indistinguishable from the HSR, and also has similarly Davies et al. (1994) for more extensive discussion]. We elevated incompatible trace elements (e.g. Y and Nb; are unable to make any further unequivocal conclusions, Table 1), suggesting that it is cogenetic with the HSR of but the physical-chemical stratification of the upper, the Mammoth Knolls. Given the chemical, isotopic, more evolved, sections of a magma chamber appears spatial and temporal similarities of the Deer Mountain the simplest and most logical geometry to explain the sample with the Mammoth Knolls we have included the observed isotope and trace element relationships. sample in the Mammoth Knoll suite. Unfortunately, no further age relationships can be When plotted in a conventional Rb–Sr isochron dia- obtained from the Rb–Sr isotope system of the post- gram (Fig. 7b), the rhyolites from Mammoth Knolls and caldera rhyolites, because of their low Rb/Sr ratios. A Deer Mountain define an isochron age of 277±124 ka U–Th study is under way in an attempt to confirm the with a mean square weighted deviation (MSWD) of 0·62. above age relationships and to determine the residence The initial ratio is 0·7063, which is indistinguishable times of the youngest (post 300 ka) lavas.

1675 JOURNAL OF PETROLOGY VOLUME 38 NUMBER 12 DECEMBER 1997

Physical evolution of the post-caldera Long to erupt. Moreover, the coherence of the chemical and Valley magma chamber thermal evolution of rhyolites (Fig. 6) with time suggests Previous geological models of the magma chamber con- one common magmatic reservoir and open-system ff figuration at Long Valley (Bailey, 1984) were based on di erentiation. the assumption that differences between early rhyolites Another significant observation is that the trace element of the Resurgent Dome and the moat rhyolites could be concentrations become significantly more variable after accounted for by the onset of cooling and crystallization ~400 ka. Before this time, trace element contents are of the Long Valley magma chamber. Later in post-caldera constant or decrease in the low-silica rhyolites and no time, differentiation processes in the upper reaches of high-silica rhyolites are produced (Figs 2 and 5). After the magma chamber led to the formation of the less 400 ka, the first high-silica rhyolites are erupted, and with time they become chemically more fractionated. voluminous high-silica moat rhyolites (Bailey, 1984) de- Given the isotopic evidence that the Mammoth Knolls rived from a dominantly low-silica volume (Hildreth, have been isolated from the main body of the magma 1981). Addition of mafic magma at the root zone of the chamber, we interpret the increased trace element vari- magmatic system was suggested to play a pivotal role in ability as evidence that sections of the magma chamber sustaining the longevity of the rhyolitic magma chamber. became spatially separated and/or stratified at about this The results from this study generally concur with the time. Subsequent magma addition did not mix with the model of Bailey (1984) for the Long Valley magma entire magma chamber. Consequently, the Sr isotope chamber, but in detail argue for a second rhyolitic melt differences between contemporaneous high-silica rhyo- being added and mixed into a residual magma reservoir. lites and low-silica rhyolites can be interpreted as a The conclusions reached in this study imply that an consequence of sampling different levels of the magma extensive crustal magma reservoir survived eruption of ff chamber, which are simultaneously changing com- the Bishop Tu and still exists beneath the Long Valley position with time. The low-silica rhyolites represent the caldera. Rhyolite genesis is a complex multi-stage process deeper and better mixed portion of the magma chamber with rhyolitic magmas formed before entering the high- whereas the high-silica rhyolites are derived from the level magma chamber, where they are extensively modi- stratified upper layer of the chamber that is gradually fied and stored. growing as the underlying chamber cools and differ- Early post-caldera rhyolites reveal marked chemical entiates following each new addition of magma. ff similarities with the late erupted Bishop Tu . The Rb–Sr The Sr and Pb isotope compositions of high-silica isotope mixing relationships defined by the Resurgent rhyolites are indistinguishable from those of preceding Dome (Table 2; Fig. 7a) provide strong evidence that, low-silica rhyolites (Fig. 5e). The Rb–Sr isochron of the ff immediately after eruption of the Bishop Tu , the magma Mammoth Knolls lavas establishes that at least some of ff chamber retained magma cogenetic to the Bishop Tu the high-silica rhyolites have been isolated for tens to eruption and that, subsequently, magmas with high Rb/ hundreds of kyr, and hence in terms of their isotope 87 86 Sr and low Sr/ Sr were added to the chamber. This compositions should be compared with older low-silica process may have been initiated when the magma par- rhyolites (Fig. 5e). Identical Sr and Pb isotopic com- ff ental to the Bishop Tu was still present in the magma positions of low- and high-silica rhyolites imply their chamber. Recent studies of melt inclusions in the lower near-contemporaneous formation. Study of the isotopic Bishop Tuff lend support to the presence of a second temporal evolution (Fig. 5e and g) requires that frac- rhyolitic magma, which probably was introduced into tionation produced the high-silica rhyolites from the main the magma chamber shortly before eruption (Lu et al., low-silica magma body at rates in excess of ~4×10–5 km3/ 1990; Dunbar & Hervig, 1992; Hervig & Dunbar, 1992). year. Unfortunately, the age resolution used to constrain Rhyolites intruded into the collapsed caldera roof these production rates is poor, such that comparison with (McConnell et al., 1995) and extrusion of Resurgent the higher production rates for pre-caldera magmas (up Dome lavas shortly after the Bishop Tuff eruptions appear to 7·5×10–4 km3/year; Davies et al., 1994) is premature to have tapped the same magma reservoir and indicate until better age resolution is obtained. The low-silica that magma survived in a chamber despite the cataclysmic rhyolites record how the majority of the magma chamber caldera-forming event. For the first ~400 kyr of its has evolved with time and provide a record of how much evolution the system produced rhyolitic magmas at an new magma has been added to the magma system. The average rate of ~0·0001 km3/year. The small volumes episodic nature of eruptions contrasts with the relatively of magmas erupted hereafter as post-caldera moat rhyo- constant temporal Sr and Pb isotope variations recorded lites do not necessarily demand a large magma chamber. by the lavas. This implies that new magmas are added However, small rhyolitic systems at depths in excess of at a relatively constant rate and that the frequency of 7 km, the inferred location of the magma chamber roof magma addition is on a shorter timescale than the (Dawson et al., 1990), are more likely to crystallize than eruption history of the entire system: i.e. <100 kyr.

1676 HEUMANN AND DAVIES POST-CALDERA RHYOLITES, LONG VALLEY

It has been argued that an increase in Nd isotope Cameron, K. L. & Cameron, M., 1986. Whole-rock/groundmass ratios before catastrophic eruptions reflects injection of differentiation trends of rare earth elements in high-silica rhyolites. large volumes of new magma into a volcanic system and Geochimica et Cosmochimica Acta 50, 759–769. Chen, C.-H., DePaolo, D. J., Nakada, S. & Shieh, Y. N., 1993. therefore represents a useful parameter in monitoring Relationship between eruption volume and neodymium isotopic the volcanic hazard potential of large high-silica rhyolite composition at Unzen volcano. Nature 362, 831–834. systems (e.g. Chen et al., 1993; DePaolo et al., 1993; Christensen, J. N. & DePaolo, D. J., 1993. Time scales of large volume DePaolo & Perry, 1997). Figure 5a clearly demonstrates silicic magma systems: Sr isotopic systematics of phenocrysts in that this is not the case for the post-caldera lavas of glass from the Bishop Tuff, Long Valley, California. Contributions to the Long Valley system, where the most voluminous Mineralogy and Petrology 113, 100–114. volcanism (75 km3) of the Resurgent Dome has the least Christensen, J. N. & Halliday, A. N., 1996. Rb–Sr ages and Nd isotopic compositions of melt inclusions from the Bishop Tuff and the radiogenic isotope ratios. The most radiogenic Nd before generation of silicic magma. Earth and Planetary Science Letters 144, the current Inyo–Mono activity occurred at ~300 ka, 547–563. 3 when <2 km of rhyolite magma was produced in a Cousens, B. L., 1992. Geochemistry and isotopic composition of post- period of 80 kyr, an eruption rate of <2·5×105 km3/ caldera basaltic lavas from Long Valley, California. EOS Transactions, year, which is below the average for the post-caldera American Geophysical Union 73, 337. magmatism. Davies, G. R. & Halliday, A. N., 1997. Development of the Long In summary, the Long Valley system represents one Valley rhyolitic magma system: Sr and Nd isotope evidence from glasses, individual phenocrysts and core–rim relationships. Geochimica of the most extensively studied silicic systems. It records et Cosmochimica Acta (in press). evidence of long-lived volcanism (>2 Myr) with several Davies, G. D., Halliday, A. N., Mahood, G. A. & Hall, C. M., 1994. periods of rapid production of high-silica rhyolites and Isotopic constraints on the production rates, crystallisation histories their subsequent storage for periods between tens of kyr and residence times of pre-caldera silicic magmas, Long Valley, and 300 kyr. California. Earth and Planetary Science Letters 125, 17–37. Dawson, P. B., Evans, J. R. & Iyer, H. M., 1990. Teleseismic tomo- graphy of the compressional wave velocity structure beneath the Long Valley region, California. Journal of Geophysical Research 95, ACKNOWLEDGEMENTS 11021–11050. DePaolo, D. J. & Perry, F. V., 1997. Nd isotopic cyclicity in rhyolite Funding for this project was provided by the Vrije of the Long Valley caldera system and implications for eruption Universiteit, Amsterdam and NWO. This manuscript forecasting. Volcanic Activity and the Environment, General Assembly IAVCEI, greatly benefited from discussions with Simone Tom- Abstracts. Puerto Vallarta: Gobierno de Jalisco, p. 111. masini and Tim Elliott, and constructive reviews by Paul DePaolo, D. J., Perry, F. V., Baldridge, W. S. & Christensen, J. N., van der Bogaard, Jake Lowenstern and Charles Dunlap. 1993. Neodymium isotopic monitor of eruption potential; postcaldera This is Publication 970146 of the Netherlands Research lavas of Long Valley, California. EOS Transactions, American Geophysical Union 4, 334. School of Sedimentary Geology. Dunbar, N. W. & Hervig, R. L., 1992. Petrogenesis and volatile stratigraphy of the Bishop Tuff: evidence from melt inclusion analysis. Journal of Geophysical Research 97, 15129–15150. Dunlap, C. E. & Gill, J. B., 1997. U-series disequilibria constraints on REFERENCES petrogenetic timescales in the large-volume, 11th century eruption Achauer, U., Greene, L., Evans, J. R. & Iyer, H. M., 1986. Nature of of Baitoushan, China/North Korea. Volcanic Activity and the Environment, the magma chamber underlying the Mono Craters area, eastern General Assembly IAVCEI, Abstracts. Puerto Vallarta: Gobierno de California, as determined from teleseismic travel time residuals. Jalisco, p. 4. Journal of Geophysical Research 91, 13873–13891. Evans, O. C. & Hanson, G. N., 1993. Accessory-mineral fractionation Arth, J. G., 1976. Behavior of trace elements during magmatic of rare-earth element (REE) abundances in granitoid rocks. Chemical processes — a summary of theoretical models and their applications. Geology 110, 69–93. Journal of Research of the US Geological Survey 4, 41–47. Gibson, R. G. & Naney, M. T., 1992. Textural development of Bailey, R. A., 1984. Chemical evolution and current state of the Long mixed, finely porphyritic silicic volcanic rocks, Inyo Domes, eastern Valley magma chamber. US Geological Survey Open File Report 84-939, California. Journal of Geophysical Research 97, 4541–4559. 24–40. Halliday, A. N., 1990. Reply to comments by R. Stephen, J. Sparks, Bailey, R. A., 1989. Geologic map of the Long Valley caldera, Mono– Herbert E. Huppert, and Colin J. N. Wilson on ‘Evidence for long Inyo-Craters volcanic chain, and vicinity, . US residence times of rhyolitic magma in the Long Valley magmatic Geological Survey Map I-1933 (1:62500). system: the isotopic record in precaldera lavas of Glass Mountain’. Bailey, R. A., Dalrymple, G. B. & Lanphere, M. A., 1976. Volcanism, Earth and Planetary Science Letters 99, 390–394. structure, and geochronology of Long Valley caldera, Mono County, Halliday, A. N., Fallick, A. E., Hutchinson, J. & Hildreth, W., 1984. California. Journal of Geophysical Research 81, 725–744. A Nd, Sr, and O isotopic investigation into the causes of chemical Baranowski, J. & Harmon, R. S., 1978. U-series chronology of two and isotopic zonation in the Bishop tuff, California. Earth and Planetary rhyolites of late Pleistocene age from Long Valley, California. In: Science Letters 68, 379–391. Zartman, R. E. (ed.) Short Papers of the Fourth International Conference, Halliday, A. N., Mahood, G. A., Holden, P., Metz, J. M., Dempster, Geochronology, Cosmochronology, Isotope Geology. US Geological Survey Open T. J. & Davidson, J. P., 1989. Evidence for long residence times of File Report 78-701, 22–24. rhyolite magma in the Long Valley magma system; the isotopic

1677 JOURNAL OF PETROLOGY VOLUME 38 NUMBER 12 DECEMBER 1997

record in the precaldera rhyolite lavas of Glass Mountain. Earth and Miller, C. D., 1985. Holocene eruptions at the Inyo volcanic chain, Planetary Science Letters 94, 274–290. California: implications for possible eruptions in Long Valley caldera. Halliday, A. N., Davidson, J. P., Hildreth, W. & Holden, P., 1991. Geology 13, 14–17. Modelling the petrogenesis of high Rb/Sr silicic magmas. Chemical Nash, W. P. & Crecraft, H. R., 1985. Partition coefficients for trace Geology 92, 107–114. elements in silicic magmas. Geochimica et Cosmochimica Acta 49, 2309– Hervig, R. L. & Dunbar, N. W., 1992. Causes of chemical zoning in 2322. the Bishop (California) and Bandelier (New Mexico) magma cham- Nielsen, D., Mahood, G. A., Halliday, A. & Essene, E. J., 1991. bers. Earth and Planetary Science Letters 111, 97–108. Petrogenesis of precaldera mafic and intermediate lavas, Long Valley, Heumann, A., Davies, G. R. & Staudigel, H., 1995. A detailed Sr–Pb CA. Geological Society of America, Abstracts with Program 23, 83. isotope study of Naivasha comendites, Kenya. In: Brown, M. & Pringle, M. S., McWilliams, M., Houghton, B. F., Lanphere, M. A. & Piccoli, P. M. (eds) US Geological Survey Circular 1129, 65. Wilson, C. J. N., 1992. 40Ar/39Ar dating of Quaternary feldspar: Higgins, M. D., 1988. Trace element geochemistry of the Inyo volcanic examples from the Taupo Volcanic Zone, New Zealand. Geology 20, chain, California; evidence for multiple magma sources, magma 531–534. mixing and post-eruption loss of boron. Journal of Volcanology and Reid, M. R., Coath, C. D., Harrison, T. M. & McKeegan, K. D., Geothermal Research 35, 97–110. 1997. Prolonged residence times for the youngest rhyolites associated Hildreth, W., 1979. The Bishop Tuff: evidence for the origin of with Long Valley Caldera: 230Th–238U ion microprobe dating of compositional zonation in silicic magma chambers. Geological Society young zircons. Earth and Planetary Science Letters 150, 27–41. of America Special Paper 180, 43–75. Sampson, D. E., 1987. Textural heterogeneities and vent area structures Hildreth, W., 1981. Gradients in silicic magma chambers: implications in the 600-year-old lavas of the Inyo volcanic chain, eastern Cali- for lithospheric magmatism. Journal of Geophysical Research 86, 10153– fornia. In: Fink, J. H. (ed.) The Emplacement of Silicic Domes and Lava 10192. Flows. Geological Society of America Special Paper 212, 89–101. Hildreth, W., 1983. Chemical differentiation of the Bishop Tuff and Sampson, D. E. & Cameron, K. L., 1987. The geochemistry of the other high-silica magmas through crystallization processes: comment. Inyo volcanic chain: multiple magma systems in the Long Valley Geology 11, 622–623. region, eastern California. Journal of Geophysical Research 92, 10403– Huppert, H. E. & Sparks, R. S. J., 1988. The generation of granitic 10421. magmas by intrusion of basalt into continental crust. Journal of Sampson, D. E., Williams, R. W., Robin, M. & Gill, J. B., 1984. The Petrology 29, 599–624. age of rhyolite magma at eruption: Inyo volcanic chain, Eastern Lu, F., Anderson, A. T., Jr & Davies, A. M., 1990. Implications of California. EOS Transactions, American Geophysical Union 65, 1128. glass inclusions for the origin of high silica rhyolite and compositional Sieh, K. & Bursik, M. I., 1986. Most recent eruption of the Mono zonation of the Bishop Tuff. EOS Transactions, American Geophysical Craters, eastern central California. Journal of Geophysical Research 91, Union 71, 651. 12539–12571. Mahood, G. A., 1990. Second reply to comment of R. S. J. Sparks, Sisson, T. W., 1994. Hornblende–melt trace-element partitioning meas- H. E. Huppert and C. J. N. Wilson on ‘Evidence for long residence ured by ion microprobe. Chemical Geology 117, 331–344. times of rhyolitic magma in the Long Valley magmatic system: and Smith, R. L., 1979. Ash-flow magmatism. Geological Society of America isotopic record in the precaldera lavas of Glass Mountain’. Earth and Special Paper 180, 5–27. Planetary Science Letters 99, 395–399. Sparks, R. S. J., Huppert, H. E. & Wilson, C. J. N., 1990. Comment Mahood, G. A. & Hildreth, W., 1983. Large partition coefficients for on ‘Evidence for long residence times of rhyolitic magma in the trace elements in high-silica rhyolites. Geochimica et Cosmochimica Acta Long Valley magmatic system: the isotopic record in precaldera 47, 11–30. lavas of Glass Mountain’ by A. N. Halliday, G. A. Mahood, P. Mankinen, E. A., Gromme, C. S., Dalrymple, G. B., Lanphere, M. A. Holden, J. M. Metz, T. J. Dempster and J. P. Davidson. Earth and & Bailey, R. A., 1986. Paleomagnetism and K–Ar ages of volcanic Planetary Science Letters 99, 387–389. rocks from Long Valley caldera, California. Journal of Geophysical Spera, F. J. & Crisp, J. A., 1981. Eruption volume, periodicity and Research 91, 633–652. caldera area: relationships and inferences on development of com- McConnell, V. S., Shearer, C. K., Eichelberger, J. C., Keskinen, M. positional zonation in silicic magma chambers. Journal of Volcanology J., Layer, P. W. & Papike, J. J., 1995. Rhyolite intrusions in the and Geothermal Research 11, 169–187. intracaldera Bishop Tuff, Long Valley Caldera, California. Journal Tanzer, M. O. & Macdougall, J. D., 1984. 230Th–238U disequilibrium of Volcanology and Geothermal Research 67, 41–60. systematics from Long Valley and Mono Domes, California; evidence Metz, J. & Mahood, G. A., 1985. Precursors to the Bishop tuff eruption: for magma replenishment since 0·1 m.y. age. EOS Transactions, Glass Mountain, Long Valley, California. Journal of Geophysical Research American Geophysical Union 65, 1128. 90, 11121–11126. Tommasini, S. & Davies, G. R., 1997. Isotope disequilibrium during Metz, J. M. & Mahood, G. A., 1991. Development of the Long Valley, anatexis: a case study of contact melting, Sierra Nevada, California. California, magma chamber recorded in precaldera rhyolite lavas, Earth and Planetary Science Letters 148, 273–286. Glass Mountain. Contributions to Mineralogy and Petrology 106, van den Bogaard, P. & Schirnick, C., 1995. 40Ar/39Ar laser probe ages 379–397. of Bishop Tuff quartz phenocrysts substantiate long-lived silicic Michael, P. J., 1983. Chemical differentiation of the Bishop Tuff and magma chamber at Long Valley, . Geology 23, 759–762. other high-silica magmas through crystallization processes. Geology Vogel, T. A., Eichelberger, J. C., Younker, L. W., Schuraytz, B. C., 11, 31–34. Horkowitz, J. P., Stockman, H. W. & Westrich, H. R., 1989. Michael, P., 1988. Partition coefficients for rare earth elements in mafic Petrology and emplacement dynamics of intrusive and extrusive minerals of high silica rhyolites: the importance of accessory mineral rhyolites of Obsidian Dome, Inyo Craters Volcanic Chain, eastern inclusions. Geochimica et Cosmochimica Acta 52, 275–282. California. Journal of Geophysical Research 94, 17937–17956.

1678